Borrowing it
Nothing to install: this file belongs to mcp-box/mcpscore. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/mcp-box/mcpscore/main/AGENTS.mdgit clone --depth 1 https://github.com/mcp-box/mcpscoreWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/instructions/mcp-box/mcpscore/agents-md)<a href="https://agentmods.dev/instructions/mcp-box/mcpscore/agents-md"><img src="https://agentmods.dev/badge/instructions/mcp-box/mcpscore/agents-md/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/instructions/mcp-box/mcpscore/agents-md"><img src="https://agentmods.dev/badge/instructions/mcp-box/mcpscore/agents-md.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.01409 | $0.01409 |
| Opus 5.5 | $0.00564 | $0.00564 |
| Sonnet 5.5 | $0.00282 | $0.00282 |
| Haiku 4.5 | $0.00141 | $0.00141 |
Grade A, and why
mcpscore AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 1mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 81 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Agent instructions for mcpscore
- Run
make allbefore handing off — it mirrors CI (lint, typecheck, tests + coverage). The gate isuv run pyright mcpscore/anduv run ruff check— inline IDE diagnostics may differ; the commands are authoritative. rule_idis a stable public contract: never rename or reuse one. New rules must cite the MCP spec section (or SEP) they enforce in their resultdetails.- Rule lifecycle — three states, and the choice between them matters:
- Active — runs everywhere.
- Version-scoped — set
min_spec_version/max_spec_versionwhen a rule is correct but only for some revisions. Use this for anything the spec removed or deprecated: the rule keeps its ID and still judges servers on the revisions where the requirement held. Obsolescence is scoping, not deletion. - Retired — delete the rule only when it was wrong (it asserted something
the spec does not require, or contradicted another rule), and add an entry to
mcpscore/rules/retired.py. That registry generates the "Retired rules" table indocs/rules.mdx, so an old report or a stale CI waiver stays explainable. A retired ID is never reused — a test enforces it.
- Readiness rules (
group_name = "readiness") score on the readiness axis; since the 1.1.0 promotion they ALSO count in the main score, but only for modern-lifecycle servers (era modern/dual-era) in full audits — never for legacy servers, never in partial audits (readiness_promotedin mcp_auditor.py is the single switch). Rules that cannot judge anything return a skip reason — they never fail a server for missing observations. - Probes must never invoke
tools/callagainst a target server — audits must be free of tool side effects. Probes never raise; network failures areProbeOutcome.ERRORdata. The ONE surface allowed to call tools is the opt-in--smokemode (smoke.py, CLI-only): its results never enter the score, it never runs from probes/rules or the web service, and its safety default (readOnlyHint-only unless--call-all) must not be weakened. docs/rules.mdxis generated — editscripts/generate_rules_doc.pyand runmake docs-rules, never the file itself. CI fails on drift.- Docs are MDX served by Mintlify from
docs/— pages need frontmatter; validate withmake docs-check. - For each new feature: update the docs and add a CHANGELOG entry (Keep a Changelog format). Docstrings: imperative first line, ≤120-char lines.
- Do not commit to
main; work on a branch. Do not add strategy or planning documents to this repository — it is public. docs-liveis a deploy pointer, never a working branch — and never delete it (normain; both are deletion-protected by repository rulesets since 2026-08-27). Mintlify serves docs.mcpscore.dev fromdocs-live;publish-docs.ymlforce-pushes it to each stable release tag. Deleting it takes the docs site down; committing to it gets overwritten at the next release. Docs changes go throughmainlike everything else and reach the site at the next release.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 1mo ago Changed · +7 lines · +71 tokens per session 4529e2abc1ed
- 2mo ago Changed · +3 lines · +70 tokens per session fd4b2d3b952d
- 2mo ago First seen · 71 lines · 1,268 tokens per session scan A 0615f9d85010
mcpscore AGENTS.md is an instructions file published in the GitHub repository mcp-box/mcpscore (42 stars, last pushed 2d ago), licensed MIT. It adds 1,409 tokens to every session, about $0.0056 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
mcpscore AGENTS.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| mcpscore AGENTS.mdmcp-box/mcpscore | Instructions file | 42 | 2d ago | MIT | A | 1,409 | Codex, OpenCode |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | today | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | yesterday | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | yesterday | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 274k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |